Heavy-storm-driven flash flood is a world-wide problem in many cities and the frequency and severity of such events have witnessed to be rapidly increasing in recent years due to climate change. To effectively predict the risk of flash floods and make appropriate adaptation planning in face of climate change, it is important to address the extreme rainfall patterns that are affected by its random nature. The project aims to develop a novel event-based nonstationary Copula method and demonstrate its applicability in assessing long-term risks of heavy storms in a tropical climate. The project involves research tasks of characterizing rainfall storms through an event-based concept, detecting the non-stationarity associated with the data time series, and carrying out univariate marginal analysis and time-varying trivariate Copula modelling for extreme storms. The method is expected to advance a new framework to evaluate the long-term risks of extreme storms, with both non-stationarity and multidimensionality been addressed. The new method proposed in this project could significantly facilitate evaluation of extreme storms in consideration of their complex nature. This is potentially helpful for water authorities in amending their management and design strategies to cope with the long-term impact of climate change.